Assessing Provectus' Pivotal Melanoma Phase 3 Trial, Part IV

The fourth in a series of blog posts and news items assessing the company's pivotal Phase 3 trial for unresectable locally advanced cutaneous melanoma (Stage III patients, and not metastatic [Stage IV] melanoma).

1. Win or lose: Will the trial succeed in meeting or fail to meet its primary and/or secondary endpoints?
2. Time-to-success or -failure: How long might it take for the trial to succeed or fail?
3. Good or bad process: Are there process steps and aspects thereof that may provide hints of potential future trial success or failure.

Entry 1: 3. Good or bad process: Patient enrollment. See Assessing Provectus' Pivotal Melanoma Phase 3 Trial, Part I (July 13, 2015) on the blog's Current News page.

Entry 2: 3. Good or bad process: Designing an interim analysis for efficacy into Provectus' pivotal melanoma Phase 3 trial. See July 16, 2015 blog post Assessing Provectus' Pivotal Melanoma Phase 3 Trial, Part II.

Entry 3: 3. Good or bad process: Patient crossover. See Assessing Provectus' Pivotal Melanoma Phase 3 Trial, Part III (July 20, 2015) on the blog's Current News page.


This blog news item, "2. Time-to-success or -failure:" Triggering the interim analysis

Blog takeaways:

• An interim analysis based on a prescribed number of events should not work for Provectus' pivotal melanoma Phase 3 trial. The trial would have to enroll 30-60% more patients than its current total patient population to reach the required number of events to trigger the interim analysis.

• Thus, the single interim analysis designed into Provectus' pivotal trial probably is based on the prescribed amount of time elapsing after which the required number of events are expected to have occurred, and

• The analysis below indicates the timing of the trial's interim analysis might occur starting in July 2016.

According to the trial protocol for the pivotal melanoma Phase 3 protocol, "[a]n interim assessment of efficacy and safety will be performed by the IRC when 50% of the events required for the primary endpoint have occurred." Call this condition the prescribed event clause. Event means disease progression — progressive disease or progression of disease ("PD").

Click to enlarge. Image source. Fuzzy purple emphasis is mine.

The trial's primary endpoint is progression-free survival ("PFS"), which is defined as "...the time from randomization until objective tumor progression or death." PD is defined as (by RECIST 1.1): "[a]t least a 20% increase in the sum of diameters of target lesions, taking as reference the smallest sum on study (this includes the baseline sum if that is the smallest on study). In addition to the relative increase of 20%, the sum must also demonstrate an absolute increase of at least 5 mm. (Note: the appearance of one or more new lesions is also considered progression)."

If one assumes one event per patient in a trial of 225 patients, where the interim analysis is undertaken when 50% of the required events (i.e., PD) occur, then ~113 events are required. That figure of 113 events, however, does not mean the trial requires 113 patients to be enrolled and treated in order to trigger the interim analysis. It means one has to wait until some number of patients are treated so that 113 events unfold in order for the interim analysis to be triggered.

A. How many patients are required to trigger the interim analysis?

Patients in both arms, the treatment group receiving PV-10 and the control one receiving chemotherapy (either dacarbazine ["DTIC"] or temozolomide ["TMZ"]), may have their disease progress, with the latter presumably generating more events faster than the former. The performance of DTIC and TMZ are well documented; see, for example, Middleton et al.: Randomized phase III study of temozolomide versus dacarbazine in the treatment of patients with advanced metastatic malignant melanoma (J Clin Oncol 2000 Jun;18(11):2351).

Click to enlarge. Middleton et al.

PV-10 performance, per patient and per lesion, from the subgroup of the company's metastatic melanoma Phase 2 trial who had all of their disease (i.e., all of their lesions) treated are below.

PV-10 Per patient

Click to enlarge. Image source. Fuzzy purple emphasis is mine.

PV-10 Per lesion

Click to enlarge. Image source. Fuzzy purple emphasis is mine.

The objective tumor responses above (in the Middleton article and on Provectus' poster presentations) generate the table and distribution curves below.

Click to enlarge.

Click to enlarge.

The two sets of distributions are somewhat mirror images of each other. Generally speaking, PV-10's CR generation is comparable in percent to chemotherapy's PD generation.

Assume chemotherapy generates disease progression events ~65-70% of the time (i.e., PD), while PV-10 generates events ~15-20% of the time. Given the trial's 2:1 randomization (2 patients receive PV-10 for 1 patient receiving chemotherapy), and assuming the aforementioned event probability ranges, the trial would not be able to achieve the required number of events (of 113) even if the trial's entire patient population were recruited, enrolled and treated. Thus, the prescribed event clause could and would not be triggered, irrespective of enrollment rate (e.g., average used to estimate the accrual period, actual determined from the trial itself).

Click to enlarge.

How many patients are required to generate the required number of events to trigger the interim analysis? About 300-360, which is greater than the trial's N of 225.

Click to enlarge.

Click to enlarge.

B. Is there another trigger that could/would generate the interim analysis?

As we saw above, particularly illustrated by the distributions above, if a substantial fraction of patients in one arm do not have their disease progress within the projected timeframe or patients in one arm fare much better than the other, then the time to accumulate the required number of events may be delayed — and, thus, the triggering of the prescribed event condition may be delayed, or not triggered at all.

A second condition would address the possibility that the triggering of first condition may be problematic: i.e., reaching the prescribed period of time after which the required number of events are expected to be occurred. Call this [second] condition the prescribed time clause. A prescribed time clause should be comprised of at least two time components: (i) the amount of time to account for enrolling and treating the prescribed number of patients to generate the required number of events, and (ii) the amount of time to account for the predicted performance of the last prescribed patient in the treatment arm.

The time to enroll and treat the prescribed number of patients. The pivotal trial's null hypothesis posits that the treatment and control arms will behave the same, or have the same response or outcome. On the high aside, assuming a PD response rate range of 65-70%, the projected number of patients that would need to be enrolled and treated is 161-173 {112.5 ÷ 0.65, 112.5 ÷ 0.7}. On the low side (i.e., SD+PD) using 85-87%, the projected number is 129-132.

The projected amount of time to recruit 129-173 patients requires the trial's projected average enrollment rate, which may be calculated by dividing the trial's N of 225 by Provectus' CTO Dr. Eric Wachter, PhD's patient accrual period of 18 months — 12.5 patients per month {225 ÷ 18}. The time to recruit 129-173 patients would be ~10-14 months {129 ÷ 12.5, 173 ÷ 12.5}.

The time for the predicted PFS performance of the last PV-10 patient. The range of this could be two or three times the projected PV-10 PFS performance (say that 10 times, quickly). PV-10's projected performance in the pivotal Phase 3 trial would be based on the trial's hazard ratio (which I believe to be 0.588, see my July 16th blog post Assessing Provectus' Pivotal Melanoma Phase 3 Trial, Part II) and systemic chemotherapy's median PFS, which in their trial Middleton et al. determined to be 1.5 months for DTIC and 1.9 months for TMZ. PV-10's projected PFS for the trial then could be: 2.6-3.2 months {1.5 ÷ 0.588, 1.9 ÷ 0.588}. Thus, the PFS performance time for the last PV-10 patient (in order to generate the interim analysis) could be: ~5-10 months {2.6 x 2, 3.2 x 3}.

Adding (a) the time to enroll/treat 113 patients to (b) the PFS performance time for the last PV-10 patient should result in (c) the prescribed amount of time to trigger the interim analysis; 10-14 months ＋ 5-10 months = 15-24 months. Given a trial commencement in April 2015, the prescribed time clause might be triggered starting in July 2016. If the enrollment is faster or slower than the projected average enrollment rate, then the prescribed time clause might be triggered sooner or later than mid-2016.

COO/CFO Peter Culpepper commented on the topic of the interim analysis in a BioTuesdays interview in May:

"Mr. Culpepper says there will be an interim data read-out when 50% of events are reached, likely in the first half of 2016. The study is expected to conclude in the third or fourth quarter of 2017. The company is targeting 25 clinical sites in the U.S., 10 in Australia, and possibly one-or-two in San Paulo and Beijing." {Underlined emphasis is mine}

Peter undoubtedly based his comments on Eric's perspective of the timing of the interim analysis. My crude analysis of said timing appears to be longer or later than Peter's timing comments.

This blog post only addresses the projection of the timing of the interim analysis. In Provectus' situation (and the topic of a future blog post or news item) reaching the minimum number of treated patients for the separation of the treatment and control arm progression-free survival curves to exceed that implied by the trial's projected hazard ratio, and reach statistical significance is more germane (and potentially a much smaller number than that required to trigger the trial's interim analysis).